There are many causes for errors. Roughly in order of importance they are:
-
Component Selection: The current sensor has low hysterisis and low tempco. The burden resistor,
calibration resistors and voltage reference all have low tempco. Offset, scaling factors and reasonable non-linearities
are not a factor for any of these devices as these errors can be calibrated out.
-
Internal Grounding: All the major sections are broken into groups and star grounded. Current paths
are strictly controlled by layout, seperate ground plane regions are used within each grounding section.
-
Separate Supplies: Supplies are provided for each section as needed. The power relay and fan control
section are totally isolated from the analog section.
-
External Grounding: An optoisloator provides galvanic isolation for the ±10V analog input for
both the MOT and Feshbach modes.
-
User Grounding: The user can only connect one ground to the system, for the power supply's voltage and
current control input. Galvanically isolation of those inputs would reduce errors further.
-
Loop Opamp Gain and FET Vgs Tempco
: This is subtle. The FETs have a temperature sensitivity on V
gs
at 125A. They are on a large heatsink a a fan to control its temperature. The temperature can be maintained between
ambient (20°) and 70° (calculated from power dissipation, airflow, heatsink theta-CA and the FETs' theta-JC).
The opamp has finite gain - the AD8067 has A
ol 119dB
typ (900,000 in real world numbers). For the opamp to adjust its output to compensate for the changing
V
gs, it has to vary its V
os. This delta-V
os
can be translated into a delta-I.
-
Ambient Temperature: Fortunately, labs are temperature controlled so the V
os
tempco of the opamp is negligible.